Comparative Study of Machine Learning Algorithms for SMS Spam Detection

Abstract

The short message service (SMS) became popular after it was initially provided as a service in the second-generation (2G) terrestrial mobile network architecture (Global System for Mobile Communication - GSM). Its popularity has been exploited by some advertising companies and others to spread unwanted advertising, communicate advertising offers, and send unwanted material to the end users. These undesirable messages, known as spam, make it difficult for the users to receive the desirable messages and make them frustration and irritation. Consequently, there are measures that various experts have implemented in filtering out these spam messages and blocking them from reaching the end users. Most of the solutions have followed the success of email spam filtering and utilized machine learning techniques to filter spam messages. The popular machine learning techniques that have successfully been used include logistical regression, Naïve Bayes algorithms, Support Vector Machine (SVM), and neural networks. The present study adopts these techniques in filtering spam messages and measures their accuracy to determine the most effective method of filtering spam messages. Based on the findings, the neural network performs best as the trained classifier model used to classify incoming messages as ham or spam.